Terraform Part 1 #42
Terraform Part 1 #42
Conversation
If `log_uri` is a "s3://..." URL, it seems to get turned into a "s3n://..." URL internally, which is different from the variable. That seems to cause terraform to want to tear down and rebuild the cluster when `terraform apply` is run with a cluster already running.
Must connect directly to the instance, the load balancer does not pass the websocket through.
| @@ -0,0 +1,5 @@ | |||
| provider "aws" { | |||
| access_key = "${var.access_key}" | |||
There was a problem hiding this comment.
This can be simplified to the snippet below if you set AWS_PROFILE when invoking terraform:
provider "aws" {
region = "${var.region}"
}| } | ||
| } | ||
|
|
||
| data "aws_instance" "emr-master" { |
| iam_role = "${aws_iam_role.ecs-service.name}" | ||
| task_definition = "${aws_ecs_task_definition.jupyterhub.arn}" | ||
|
|
||
| load_balancer { |
There was a problem hiding this comment.
You weren't able to omit this? Does it work via the ELB now or is it still broken because of the WebSocket dependency?
There was a problem hiding this comment.
I have not attempted to, yet. I would like to merge this as-is in order for the image changes to be published, then I will come back and implement your suggestions.
So far my recollections are: (1) splitting up of the security groups and (2) removing the load_balancer block. Are there any others?
There was a problem hiding this comment.
Yeah, security_group_rule and trying to remove the load_balancer block sounds good.
There was a problem hiding this comment.
I have a work in progress PR up do deal with those two issues: #45.
To test:
quay.io).terraformdirectoryterraform initterraform applyterraform destroyStill needs: